CSUN professor warns AI’s cheaper campaign ads come with a trust problem
In a newly published interview, Kristen Walker describes a tension between affordable campaign tools and increasingly convincing fabrication.
Listen to this story
The audio brief
Story brief
3 key pointsKristen Walker’s warning puts a practical trade-off at the center of AI political advertising: cheaper production can help grassroots-funded candidates compete, while synthetic or AI-enhanced content makes truthful representation harder to judge. A Wesleyan Media Project snapshot identified at least 164 AI-generated or enhanced ads by September 4, tied to just under $80 million in estimated media spending—not AI-tool costs—and researchers said the count likely missed hard-to-detect uses. With 31 states regulating political deepfakes or requiring disclosure, campaigns and regulators are testing where the line should be drawn.
- 01
In the identified sample, 69% of ads had no AI disclaimer, and undisclosed ads accounted for 78% of estimated spending.
- 02
Republican candidates and pro-Republican groups represented 80% of ads and 83% of spending; pro-Democratic sponsors accounted for 11% and 14%, respectively.
- 03
The tally covered broadcast television, connected TV, and digital ads, and included AI-enhanced material as well as fully generated content.
Campaigns with little money can now make attractive political ads with AI, but that access brings a harder question about truthfulness, California State University, Northridge marketing professor Kristen Walker says. In a November 18 interview published by CSUN Newsroom, she warned that the boundary between authentic and fabricated political advertising is becoming harder to see.
Walker, who studies technology and data privacy at CSUN’s David Nazarian College of Business and Economics, described uses ranging from flyers and infographics to synthetic videos and deepfakes of candidates. The tools are not limited to inventing an entire scene: campaigns can use them across different kinds of campaign material.
Her argument was not simply that campaigns should avoid the technology. She said affordable production could help candidates relying on grassroots funding compete with better-resourced opponents. Access to polished marketing tools, in her view, offers an advantage that those candidates might otherwise lack. She paired that benefit with an explicit hope that campaigns would use it ethically.
“They might be using it because it’s more affordable and it gives them a lot of marketing power. But are they being truthful?” Walker asked. That distinction separates her endorsement of wider access from approval of whatever a campaign chooses to produce with it.
An earlier tally shows the disclosure gap
The interview points back to a September 4 Wesleyan Media Project release. By that date, researchers had identified at least 164 political ads generated or enhanced with AI, with estimated spending just under $80 million. Those figures are a dated snapshot, not a final accounting of the midterm cycle.
The spending estimates came from AdImpact and concerned the identified advertisements across broadcast television, connected TV and digital venues. They were not a measure of what campaigns paid for AI tools. The category also included AI-enhanced ads, rather than only fully generated videos or fabricated depictions of candidates.
Ads lacking disclosure accounted for 78% of the spending in the project’s identified sample.
Researchers assembled the list through news coverage, searches for AI disclaimers and their own tracking of visuals and deepfakes—videos depicting someone doing or saying something they did not do. Suspected uses flagged by coding staff were confirmed by the team. Wesleyan warned that difficult-to-detect AI use made its tally an underestimate.
The identified activity was heavily concentrated among Republican sponsors. Republican candidates and pro-Republican groups accounted for 80% of the ads and 83% of spending in that September sample. Pro-Democratic sponsors accounted for 11% of ads and 14% of spending; other sponsors made up the remainder.
Rules meet campaigns testing the boundaries
CSUN’s November account says 31 states have laws prohibiting political deepfakes or requiring disclosures, citing the National Conference of State Legislatures. It also says the Supreme Court has yet to take up a case addressing those laws’ constitutionality. Walker described campaigns as testing the boundaries of regulations and existing consumer protections.
Walker expects more regulation, but she also anticipates that distinguishing facts from fabrication will become harder. She framed the next step as a public judgment about acceptable uses, not only a technical challenge: “I think as a society we need to ask ourselves what do we think is okay; how much of it is okay; and when is it okay?”
Sources
- mediaproject.wesleyan.eduAt Least $80 Million Spent on AI Ads This Cycle - Wesleyan Media Project
- newsroom.csun.eduReal or AI-Generated? Political Ads Face a New Reality, CSUN Prof Says - CSUN Newsroom
Reader comments
Newest comments first. Replies stay oldest first.